Straightforward, functional installation on Apple Silicon // moving to PySide6 // rc2 by jeylau · Pull Request #2022 · DeepLabCut/DeepLabCut

jeylau · 2022-10-22T09:24:24Z

Transition from pyside2 to pyside6–only the latter is supported on M1/M2 chips (https://wiki.qt.io/Qt_for_Python)
Utilize Apple's tensorflow plugin (https://developer.apple.com/metal/tensorflow-plugin/). Note that I had to remove tensorflow from the required dependencies in the setup.py; we otherwise run into an error due to the mismatch between tensorflow and tensorflow-macos. I would argue that's also advantageous, as this allows for lighter installation e.g. for users training remotely, or lets users pip install the version matching their GPU device.

I paste below the simple steps to follow to get a working installation with python 3.9 and tensorflow 2.10:

We'll get the miniconda M1 bash installer, as explained in https://docs.conda.io/projects/conda/en/latest/user-guide/install/macos.html. In the Terminal, run the following commands:

wget https://repo.anaconda.com/miniconda/Miniconda3-py39_4.12.0-MacOSX-arm64.sh -O ~/miniconda.sh
bash ~/miniconda.sh -b -p $HOME/miniconda
source ~/miniconda/bin/activate
conda init zsh

Then, clone DeepLabCut (git clone https://github.com/DeepLabCut/DeepLabCut.git), and run conda env create -f conda-environments/DEEPLABCUT_M1.yaml. Next, activate the environment (conda activate DEEPLABCUT_M1), and launch DLC with pythonw -m deeplabcut.

MMathisLab · 2022-10-29T12:22:47Z

bringing in setup.py changes from PR #1970

MMathisLab

I found the miniconda install annoying, as it broke all my anaconda env, so that is a warning to users...

MMathisLab · 2022-10-29T12:54:34Z

EDIT: resolved!

@jeylau please address:

Testing on macbook intel chip (2020), v12.5.1:

git checkout m1m2

pip install -e '.[gui,tf]'

python -m deeplabcut

GUI launches, load demo openfield project,

Loading DLC 2.3rc1...
Starting GUI...
/Users/mwmathis/opt/anaconda3/envs/DEEPLABCUT/lib/python3.10/site-packages/qtpy/__init__.py:267: RuntimeWarning: Selected binding "pyside2" could not be found, using "pyside6"
  warnings.warn('Selected binding "{}" could not be found, '

edit config to extract 2 frames, go to extract, and it hangs on yes/no, which I cannot select ...

have to hard crash out of it :(.

Traceback (most recent call last):
  File "/Users/mwmathis/Documents/DeepLabCut/deeplabcut/gui/tabs/label_frames.py", line 31, in label_frames
    dialog.setFileMode(dialog.DirectoryOnly)
AttributeError: 'PySide6.QtWidgets.QFileDialog' object has no attribute 'DirectoryOnly'. Did you mean: 'directoryUrl'?

jeylau · 2022-10-31T14:51:43Z

I found the miniconda install annoying, as it broke all my anaconda env, so that is a warning to users...

Yeah I agree...
Perhaps we could get anaconda arm64 from the bottom of that page, I'm not sure whether that would avoid breaking conda though.

MMathisLab · 2022-11-02T16:48:43Z

Install now works great for me! Note, it does require miniconda3, as normal anaconda does fail ;) I had to check :)

but, when I run the python testscript.py it is missing tf_slim:

Loading DLC 2.3rc1...
DLC loaded in light mode; you cannot use any GUI (labeling, relabeling and standalone GUI)
  File "/Users/mwmathis/Documents/DeepLabCut/deeplabcut/pose_estimation_tensorflow/core/train.py", line 24, in <module>
    import tf_slim as slim
ModuleNotFoundError: No module named 'tf_slim'

so I need to run:
pip install tf_slim

then it runs 👍

MMathisLab · 2022-11-02T16:51:10Z

but it seems to not download the model correctly:

❯ python testscript.py
Loading DLC 2.3rc1...
Imported DLC!
On Windows/OSX tensorpack is not tested by default.
CREATING PROJECT
Created "/Users/mwmathis/Documents/DeepLabCut/examples/TEST-Alex-2022-11-02/videos"
Created "/Users/mwmathis/Documents/DeepLabCut/examples/TEST-Alex-2022-11-02/labeled-data"
Created "/Users/mwmathis/Documents/DeepLabCut/examples/TEST-Alex-2022-11-02/training-datasets"
Created "/Users/mwmathis/Documents/DeepLabCut/examples/TEST-Alex-2022-11-02/dlc-models"
Copying the videos
/Users/mwmathis/Documents/DeepLabCut/examples/TEST-Alex-2022-11-02/videos/reachingvideo1.avi
Generated "/Users/mwmathis/Documents/DeepLabCut/examples/TEST-Alex-2022-11-02/config.yaml"

A new project with name TEST-Alex-2022-11-02 is created at /Users/mwmathis/Documents/DeepLabCut/examples and a configurable file (config.yaml) is stored there. Change the parameters in this file to adapt to your project's needs.
 Once you have changed the configuration file, use the function 'extract_frames' to select frames for labeling.
. [OPTIONAL] Use the function 'add_new_videos' to add new videos to your project (at any stage).
EXTRACTING FRAMES
Config file read successfully.
Extracting frames based on kmeans ...
Kmeans-quantization based extracting of frames from 0.0  seconds to 8.53  seconds.
Extracting and downsampling... 256  frames from the video.
256it [00:01, 176.07it/s]
Kmeans clustering ... (this might take a while)
Frames were successfully extracted, for the videos listed in the config.yaml file.

You can now label the frames using the function 'label_frames' (Note, you should label frames extracted from diverse videos (and many videos; we do not recommend training on single videos!)).
CREATING-SOME LABELS FOR THE FRAMES
Plot labels...
Creating images with labels by Alex.
100%|██████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 5/5 [00:00<00:00,  5.70it/s]
If all the labels are ok, then use the function 'create_training_dataset' to create the training dataset!
CREATING TRAININGSET
The training dataset is successfully created. Use the function 'train_network' to start training. Happy training!
CHANGING training parameters to end quickly!
TRAIN
Selecting single-animal trainer
Config:
{'all_joints': [[0], [1], [2], [3]],
 'all_joints_names': ['bodypart1', 'bodypart2', 'bodypart3', 'objectA'],
 'alpha_r': 0.02,
 'apply_prob': 0.5,
 'batch_size': 1,
 'contrast': {'clahe': True,
              'claheratio': 0.1,
              'histeq': True,
              'histeqratio': 0.1},
 'convolution': {'edge': False,
                 'emboss': {'alpha': [0.0, 1.0], 'strength': [0.5, 1.5]},
                 'embossratio': 0.1,
                 'sharpen': False,
                 'sharpenratio': 0.3},
 'crop_pad': 0,
 'cropratio': 0.4,
 'dataset': 'training-datasets/iteration-0/UnaugmentedDataSet_TESTNov2/TEST_Alex80shuffle1.mat',
 'dataset_type': 'default',
 'decay_steps': 30000,
 'deterministic': False,
 'display_iters': 2,
 'fg_fraction': 0.25,
 'global_scale': 0.8,
 'init_weights': '/Users/mwmathis/Documents/DeepLabCut/deeplabcut/pose_estimation_tensorflow/models/pretrained/resnet_v1_50.ckpt',
 'intermediate_supervision': False,
 'intermediate_supervision_layer': 12,
 'location_refinement': True,
 'locref_huber_loss': True,
 'locref_loss_weight': 0.05,
 'locref_stdev': 7.2801,
 'log_dir': 'log',
 'lr_init': 0.0005,
 'max_input_size': 1500,
 'mean_pixel': [123.68, 116.779, 103.939],
 'metadataset': 'training-datasets/iteration-0/UnaugmentedDataSet_TESTNov2/Documentation_data-TEST_80shuffle1.pickle',
 'min_input_size': 64,
 'mirror': False,
 'multi_stage': False,
 'multi_step': [[0.001, 5]],
 'net_type': 'resnet_50',
 'num_joints': 4,
 'optimizer': 'sgd',
 'pairwise_huber_loss': False,
 'pairwise_predict': False,
 'partaffinityfield_predict': False,
 'pos_dist_thresh': 17,
 'project_path': '/Users/mwmathis/Documents/DeepLabCut/examples/TEST-Alex-2022-11-02',
 'regularize': False,
 'rotation': 25,
 'rotratio': 0.4,
 'save_iters': 5,
 'scale_jitter_lo': 0.5,
 'scale_jitter_up': 1.25,
 'scoremap_dir': 'test',
 'shuffle': True,
 'snapshot_prefix': '/Users/mwmathis/Documents/DeepLabCut/examples/TEST-Alex-2022-11-02/dlc-models/iteration-0/TESTNov2-trainset80shuffle1/train/snapshot',
 'stride': 8.0,
 'weigh_negatives': False,
 'weigh_only_present_joints': False,
 'weigh_part_predictions': False,
 'weight_decay': 0.0001}
Batch Size is 1
/Users/mwmathis/miniconda/envs/DEEPLABCUT_M1/lib/python3.9/site-packages/tensorflow/python/keras/engine/base_layer_v1.py:1694: UserWarning: `layer.apply` is deprecated and will be removed in a future version. Please use `layer.__call__` method instead.
  warnings.warn('`layer.apply` is deprecated and '
Metal device set to: Apple M1 Pro

systemMemory: 32.00 GB
maxCacheSize: 10.67 GB

2022-11-02 17:49:29.132517: I tensorflow/core/common_runtime/pluggable_device/pluggable_device_factory.cc:306] Could not identify NUMA node of platform GPU ID 0, defaulting to 0. Your kernel may not have been built with NUMA support.
2022-11-02 17:49:29.132647: I tensorflow/core/common_runtime/pluggable_device/pluggable_device_factory.cc:272] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 0 MB memory) -> physical PluggableDevice (device: 0, name: METAL, pci bus id: <undefined>)
Loading ImageNet-pretrained resnet_50
2022-11-02 17:49:29.302387: I tensorflow/core/common_runtime/pluggable_device/pluggable_device_factory.cc:306] Could not identify NUMA node of platform GPU ID 0, defaulting to 0. Your kernel may not have been built with NUMA support.
2022-11-02 17:49:29.302418: I tensorflow/core/common_runtime/pluggable_device/pluggable_device_factory.cc:272] Created TensorFlow device (/job:localhost/replica:0/task:0/device:GPU:0 with 0 MB memory) -> physical PluggableDevice (device: 0, name: METAL, pci bus id: <undefined>)
/Users/mwmathis/Documents/DeepLabCut/deeplabcut/pose_estimation_tensorflow/core/train.py:225: UserWarning: Switching to Adam, as SGD crashes on Apple Silicon.
  warnings.warn("Switching to Adam, as SGD crashes on Apple Silicon.")
2022-11-02 17:49:29.701147: I tensorflow/compiler/mlir/mlir_graph_optimization_pass.cc:354] MLIR V1 optimization pass is not enabled
2022-11-02 17:49:29.726137: W tensorflow/core/platform/profile_utils/cpu_utils.cc:128] Failed to get CPU frequency: 0 Hz
2022-11-02 17:49:29.731388: I tensorflow/core/grappler/optimizers/custom_graph_optimizer_registry.cc:114] Plugin optimizer for device_type GPU is enabled.
2022-11-02 17:49:30.449692: I tensorflow/core/grappler/optimizers/custom_graph_optimizer_registry.cc:114] Plugin optimizer for device_type GPU is enabled.
2022-11-02 17:49:31.673542: I tensorflow/core/grappler/optimizers/custom_graph_optimizer_registry.cc:114] Plugin optimizer for device_type GPU is enabled.
Traceback (most recent call last):
  File "/Users/mwmathis/Documents/DeepLabCut/examples/testscript.py", line 164, in <module>
    deeplabcut.train_network(path_config_file)
  File "/Users/mwmathis/Documents/DeepLabCut/deeplabcut/pose_estimation_tensorflow/training.py", line 221, in train_network
    raise e
  File "/Users/mwmathis/Documents/DeepLabCut/deeplabcut/pose_estimation_tensorflow/training.py", line 210, in train_network
    train(
  File "/Users/mwmathis/Documents/DeepLabCut/deeplabcut/pose_estimation_tensorflow/core/train.py", line 242, in train
    restorer.restore(sess, cfg["init_weights"])
  File "/Users/mwmathis/miniconda/envs/DEEPLABCUT_M1/lib/python3.9/site-packages/tensorflow/python/training/saver.py", line 1407, in restore
    raise ValueError("The passed save_path is not a valid checkpoint: " +
ValueError: The passed save_path is not a valid checkpoint: /Users/mwmathis/Documents/DeepLabCut/deeplabcut/pose_estimation_tensorflow/models/pretrained/resnet_v1_50.ckpt

jeylau · 2022-11-02T18:26:38Z

@MMathisLab, was this run on a M1 or M2 chip? (Although I'm not sure it makes a big difference)

MMathisLab · 2022-11-04T13:43:17Z

alright, ALL solved! easy install, great work @jeylau :)

* Fix typo * Fix failing tests (#2021) * Fix pandas regression * Update requirements.txt * Update installation.md (#2020) * Update installation.md * Update installTips.md * Update installation.md * Add benchmark API (#1777) * Add benchmark API Co-authored-by: Jessy <30733203+jeylau@users.noreply.github.com> * black formatting, added ref for datasets in benchmark * Update installation.md (#2029) Just a typo Co-authored-by: Jessy Lauer <30733203+jeylau@users.noreply.github.com> Co-authored-by: Steffen Schneider <steffen@bethgelab.org> Co-authored-by: AlexEMG <alexander@deeplabcut.org> Co-authored-by: Ale Santuz <alessandro.santuz@gmail.com>

…o PySide6 // rc2 (DeepLabCut#2022) * Drop pyside2 for pyside6 * Functional Apple Silicon conda environment file

jeylau added 3 commits October 22, 2022 00:40

Drop pyside2 for pyside6

43fa94a

Functional Apple Silicon conda environment file

56e59b1

Fix pyside6 bugs

749a96a

jeylau added the Installation label Oct 22, 2022

jeylau requested a review from MMathisLab October 22, 2022 09:24

jeylau self-assigned this Oct 22, 2022

jeylau mentioned this pull request Oct 26, 2022

DLC installation errors on Mac M2 #2011

Closed

2 tasks

Update setup.py

84dbab1

MMathisLab approved these changes Oct 29, 2022

View reviewed changes

typo fix

a769a4d

MMathisLab self-requested a review October 29, 2022 12:40

MMathisLab changed the title ~~Straightforward, functional installation on Apple Silicon~~ Straightforward, functional installation on Apple Silicon // moving to PySide6 Oct 29, 2022

jeylau added 6 commits October 31, 2022 13:07

No feedback when extracting frames

3c94de5

Update setup.py

01ff18f

Exit pyside qthread gracefully

83bf78c

Revert git install in setup.py

f8de24e

Fix setFileMode

cd9ed6a

Fix file dialog crash

bea96c0

jeylau added 2 commits October 31, 2022 16:21

Fix qtpy RuntimeWarning upon GUI launch

5f04465

Use Adam rather than SGD on Apple Silicon

6898b99

jeylau marked this pull request as draft November 2, 2022 16:17

Delete testM1.sh

d3309ec

MMathisLab marked this pull request as ready for review November 4, 2022 13:10

Update installation.md

c6e7843

MMathisLab added 5 commits November 4, 2022 14:33

keep tf_slim

6ce6c2b

Update setup.py

380395d

rename to tf as added install...

2aa83f7

Update installation.md

ce76446

Update installation.md

777dae4

MMathisLab approved these changes Nov 4, 2022

View reviewed changes

MMathisLab and others added 5 commits November 4, 2022 14:47

Update DEEPLABCUT.yaml

9ec6c75

rc2

0587d0c

Update version.py

44b6b86

Update setup.py

805a264

MMathisLab mentioned this pull request Nov 4, 2022

Mwm/apple m1 [WIP] #1970

Closed

MMathisLab changed the title ~~Straightforward, functional installation on Apple Silicon // moving to PySide6~~ Straightforward, functional installation on Apple Silicon // moving to PySide6 // rc2 Nov 4, 2022

jeylau added 2 commits November 4, 2022 17:12

Fix project loading dialog

63a915e

Fix config IO in ConfigEditor

3aca7c9

MMathisLab approved these changes Nov 4, 2022

View reviewed changes

AlexEMG approved these changes Nov 4, 2022

View reviewed changes

minor versioning

0212f6a

AlexEMG merged commit 993a640 into master Nov 4, 2022

jeylau deleted the m1m2 branch November 8, 2022 08:44

n-poulsen mentioned this pull request May 5, 2023

removed deprecated install tips for mac m1/m2 #2234

Merged

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

Straightforward, functional installation on Apple Silicon // moving to PySide6 // rc2#2022

Straightforward, functional installation on Apple Silicon // moving to PySide6 // rc2#2022
AlexEMG merged 28 commits intomasterfrom
m1m2

jeylau commented Oct 22, 2022

Uh oh!

MMathisLab commented Oct 29, 2022

Uh oh!

MMathisLab left a comment

Uh oh!

MMathisLab commented Oct 29, 2022 •

edited

Loading

Uh oh!

jeylau commented Oct 31, 2022 •

edited

Loading

Uh oh!

MMathisLab commented Nov 2, 2022 •

edited

Loading

Uh oh!

MMathisLab commented Nov 2, 2022

Uh oh!

jeylau commented Nov 2, 2022

Uh oh!

MMathisLab commented Nov 4, 2022

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Uh oh!

Conversation

jeylau commented Oct 22, 2022

Uh oh!

MMathisLab commented Oct 29, 2022

Uh oh!

MMathisLab left a comment

Choose a reason for hiding this comment

Uh oh!

MMathisLab commented Oct 29, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

jeylau commented Oct 31, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

MMathisLab commented Nov 2, 2022 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

MMathisLab commented Nov 2, 2022

Uh oh!

jeylau commented Nov 2, 2022

Uh oh!

MMathisLab commented Nov 4, 2022

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

MMathisLab commented Oct 29, 2022 •

edited

Loading

jeylau commented Oct 31, 2022 •

edited

Loading

MMathisLab commented Nov 2, 2022 •

edited

Loading